A phone number or email address can help find a customer, but it is not always a stable identity. People share addresses, change numbers, use aliases, and appear in several organisations' systems.
Use an application-controlled customer identity and explicit mappings to external records. Keep lookup evidence separate from the decision to create, update, or merge a customer.
Scope mappings to the external account
Record the provider, connected account or tenant, external record identifier, and internal customer reference. Two clients using the same provider can legitimately have the same-looking external identifier in different account contexts.
Define which fields each system owns and how changes propagate. A CRM contact identifier can identify a provider record without proving that every incoming enquiry belongs to the same real-world person.
Resolve a fictional repeat-enquiry scenario
Suppose a customer contacts a business with a new phone number but an existing email. Another family member uses the same email for a separate service address. A naive email match could merge their histories incorrectly.
Create fixtures for exact matches, changed contact details, shared addresses, conflicting matches, and no match. For each, define whether the system links automatically, creates a new record, or requests human review. Record the evidence used for that decision.
Then connect two fictional client organisations to the same sales provider. Test that a lookup or mapping from one client never resolves to the other's customer merely because an external ID or email matches. Inspect exports and support tools as well as the intake path.
Make merges and corrections reversible where practical
An identity mistake can spread through integrations. Keep enough mapping history to understand which records were joined and where updates were sent. Define the process for correcting a mistaken association rather than treating merge as an invisible database cleanup.
Repeated events should resolve to the same intended mapping under the chosen duplicate policy. Delayed events for a deleted or merged customer need an explicit outcome. Do not let an old provider payload silently recreate a removed identity without review.
The tradeoff is automation versus false association. Aggressive matching reduces duplicate records but can combine unrelated people. Conservative matching creates more review work. Choose thresholds and required evidence from the workflow's consequences, then evaluate representative fixtures instead of assuming one contact field is universally unique.
The HVAC-Automate portfolio account describes a customer and service-request workflow. The identity scenarios here are fictional and do not assert a particular deployed matching algorithm.
Questions about customer identifiers
Can email be a permanent primary identity?
It may be a useful lookup field, but changes, sharing, and provider differences require a deliberate policy. Separate stable internal identity from mutable contact data.
Should every duplicate be merged automatically?
No. Ambiguous matches may need review, especially when histories, payments, or private records could be combined.
What should an integration review inspect?
Inspect mapping scope, changed contacts, shared fields, merge history, and delayed events. Bring a small fictional identity matrix to a data-integration discussion.